Three basic stages towards the global modeling of the eye are presented. In the first stage, an adequate choice of the basis geometrical model, general ellipsoid in this case, permits, to fit in a natural way the typical "melon" shape of the cornea with minimum complexity. In addition it facilitates to extract most of its optically relevant parameters, such as the position and orientation of it optical axis in the 3D space, the paraxial and overall refractive power, the amount and axis of astigmatism, etc. In the second level, this geometrical model, along with optical design and optimization tools, is applied to build customized optical models of individual eyes, able to reproduce the measured wave aberration with high fidelity. Finally, we put together a sequence of schematic, but functionally realistic models of the different stages of image acquisition, coding and analysis in the visual system, along with a probabilistic Bayesian maximum a posteriori identification approach. This permitted us to build a realistic simulation of the all the essential processes involved in a visual acuity clinical exam. It is remarkable that at all three levels, it has been possible for the models to predict the experimental data with high accuracy.